The Monitor tick ran smartctl against every physical disk every 5
seconds, regardless of which tab was visible — and smartctl's default
behaviour is to spin up a drive in standby to read its attributes. Net
effect: drives could never spin down and ran (hot) 24/7 while the TUI
was open.
Three fixes:
- smartctl now runs with -n standby: a sleeping drive is never woken
for a health probe. It's reported as a new Standby state and shown
as "asleep" in the SMART table — a good sign, not a missing reading.
- Health probes only run while the Monitor tab is visible. The tick
always re-arms, but Setup/Manage now cost zero disk activity.
Switching to Monitor (or pressing r on it) fires an immediate
refresh so the snapshot is never stale.
- SMART moved to its own 60s cadence. The 5s tick keeps the cheap
probes (statfs, /proc, systemd); disk rows are cached between SMART
probes so the table doesn't flicker. lastSmart is stamped at issue
time so a slow USB bridge can't double up probes.
Verified on moviebox (3x 10TB spinning drives) with a counting shim
around smartctl: 0 calls while on Setup, immediate probe on entering
Monitor, then exactly one batch per 60s.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
28allday
262ba8e3b4